rbpred

RB-Pred: Support Vector Machine-Based Rice Blast Severity Prediction

RB-Pred predicts rice blast severity using machine learning models trained on weather variables to quantify plant–pathogen–environment interactions.

Key Features:

• Support Vector Machine (SVM) Modeling: Implements SVM algorithms for disease severity prediction, outperforming neural networks and multiple regression models in correlation coefficient (r) and percent mean absolute error (%MAE).
• Weather-Based Predictors: Utilizes six significant weather variables as input features to model rice blast disease dynamics.
• Five-Fold Cross-Validation: Applies cross-location and cross-year validation to assess model robustness and generalizability.

Scientific Applications:

• Rice Blast Forecasting: Quantifies disease severity to support epidemiological analysis and optimization of control measure timing.
• Plant Disease Epidemiology Research: Enables investigation of weather-driven pathogen dynamics and plant–pathogen–environment interactions.

Methodology:

Six weather variables influencing rice blast occurrence were selected as predictors. Models were trained using a support vector machine algorithm and evaluated through five-fold cross-validation, including cross-location and cross-year analyses. Performance was assessed using correlation coefficient (r) and percent mean absolute error (%MAE).